RegTraffic: A Regression Based Traffic Simulator for Spatiotemporal Traffic Modeling, Simulation and Visualization
This work addresses the need for more interactive traffic simulators for urban planning and management, though it appears incremental as it builds on existing regression methods for traffic analysis.
The paper tackles the problem of interactive and adaptive traffic modeling by introducing RegTraffic, a novel simulator that uses dynamic regression-based spatiotemporal analysis to predict congestion on correlated road segments, achieving a Mean Squared Error of 1.3 Km/h and Root Mean Squared Error of 1.71 Km/h.
Traffic simulation is a great tool to demonstrate complex traffic structures which can be extremely useful for the planning, development, and management of road traffic networks. Current traffic simulators offer limited features when it comes to interactive and adaptive traffic modeling. This paper presents RegTraffic, a novel interactive traffic simulator that integrates dynamic regression-based spatiotemporal traffic analysis to predict congestion of intercorrelated road segments. The simulator models traffic congestion of road segments depending on neighboring road links and temporal features of the dynamic traffic flow. The simulator provides a user-friendly web interface to select road segments of interest, receive user-defined traffic parameters, and visualize the traffic for the flow of correlated road links based on the user inputs and the underlying correlation of these road links. Performance evaluation shows that RegTraffic can effectively predict traffic congestion with a Mean Squared Error of 1.3 Km/h and a Root Mean Squared Error of 1.71 Km/h. RegTraffic can effectively simulate the results and provide visualization on interactive geographical maps.